The dengue viruses are members of the family Flaviviridae which also includes the Japanese encephalitis (JE) virus, Tick-borne encephalitis (TBE) virus, and the is initially discovered prototype of this class, the yellow fever (YF) virus. The flaviviruses contain a single positive strand genomic RNA and are small enveloped viruses affecting animals, but generally transmitted to vertebrates by chronically infected mosquito or tick vectors. Flaviviruses are enveloped by host cell membrane and contain the three structural proteins capsid (C), membrane (M), and envelope (E). The E and M proteins are found on the surface of the virion where they are anchored in the membrane. Mature E is glycosylated, whereas M is not, although its precursor, preM, is a glycoprotein. Glycoprotein E, the largest structural protein, contains functional domains responsible for cell surface attachment and intraendosomal fusion activities. It is also a major target of the host immune system, inducing neutralizing antibodies, protective immunity, as well as antibodies which inhibit hemagglutination.
Dengue virus is the causative agent of dengue fever and is transmitted to man by Aedes mosquitoes, principally Aedes aegypti and Aedes albopictus. Classic dengue fever is an acute illness marked by fever, headache, aching muscles and joints, and rash. A fraction of cases, typically in children, results in more extreme forms of infection, i.e., dengue hemorrhagic fever (DHF) or dengue shock syndrome (DSS). Without diagnosis and prompt medical intervention, the sudden onset and rapid progress of DHF/DSS can be fatal.
Dengue is one of the most important virus groups transmitted to man by arthropods in terms of global morbidity; it has been estimated that dengue is responsible for up to 100 million illnesses annually. With the advent of modern jet travel, dengue has spread globally in the tropics and subtropics, and multiple dengue serotypes in a region are common.
Every flavivirus genome is a single positive-stranded RNA of approximately 10,500 nucleotides containing short 5' and 3' untranslated regions, a single long open reading frame (ORF), a 5' cap, and a nonpolyadenylated 3' terminus. The ten gene products encoded by the single, long open reading frame are contained in a polyprotein organized in the order, C (capsid), preM/M (membrane), E (envelope), NS1 (nonstructural), NS2a, NS2b, NS3, NS4a, NS4b, and NS5 (Chambers, T. J. et al. Ann Rev Microbiol (1990) 44:649-688). Processing of the encoded polyprotein is initiated cotranslationally, and full maturation requires both host and viral-specific proteases. The sites of proteolytic cleavage in the YF virus have been determined by comparing the nucleotide sequence and the amino terminal sequences of the viral proteins. Subsequent to initial processing of the polyprotein, preM is converted to M during virus release (Wengler, G. et al. J Virol (1989) 63:2521-2526), and anchored C is processed during virus maturation (Nowak et al. Virology (1987) 156:127-137).
There are four antigenically related dengue viruses which, however, can be recognized as distinct serotypes. The complete genomic sequence for at least one strain of each of the four dengue serotypes has been reported (DEN-1, Fu, J. et al. Virology (1992) 188:953-958; DEN-2, Deubel, V. et al. Virology (1986) 155:365-377; Hahn, Y. S. et al. Virology (1988) 162:167-180; DEN-3, Osatomi, K. et al. Virus Genes (1988) 2:99-108; Osatomi, K. et al. Virology (1990) 176:643-647; DEN-4, Zhao, B. E. et al. Virology (1986) 155:77-88; Mackow, E. et al. Virology (1987) 159:217-228). In addition, the compete genomic sequences of other flaviviruses are known (e.g., YF virus: Rice et al., Science (1985) 229:726-733).
It does not appear that infection by one dengue serotype can confer long-term immunity on the individual with respect to other serotypes. In fact, secondary infections with heterologous serotypes are quite common. In general, antibody responses in infected subjects to primary infection are mostly IgM antibodies and these antibodies are directed against type-specific determinants. On the other hand, secondary infections by heterologous serotypes generate IgG antibodies which are flavivirus crossreactive.
Helpful reviews of the nature of the dengue disease, the history of attempts to develop suitable vaccines, and structural features of flaviviruses in general as well as the molecular structural features of the envelope protein of flaviviruses are found in Halstead, S. B. Science (1988) 239:476-481; Brandt, W. E. J Infect Disease (1990) 162:577-583; Chambers, T. J. et al. Annual Rev Microbiol (1990) 44:649-688; Mandl, C. W. et al. Virology (1989) 63:564-571; and Henchal, E. A. and J. R. Putnak, Clin Microbiol Rev (1990) 3:376-396.
A successful vaccine for protection against dengue infection has never been developed. However, there have been a number of preliminary efforts, many of which focus on the envelope protein, since this protein is exposed at the surface and is believed to be responsible for eliciting immunity.
Monoclonal antibodies (Mabs) directed against purified E of several flaviviruses DEN-2 (Henchal et al. Am J Trop Med Hyg (1985) 34:162-169, TBE (Heinz, F. X. et al. Virology (1983) 126:525-537), St. Louis encephalitis (SLE, Mathews, J. H. et al. J Immunol (1984) 132:1533-1537), Murray Valley encephalitis (MVE, Hawkes, R. A. et al. J Gen Virol (1988) 69:1105-1109), and JE (Takegami, T. et al. Acta Virologica (1982) 26:312-320) are neutralizing in vitro. Some of these Mabs can also passively confer protection in vivo (Heinz, F. X. et al. (1983, supra)); Mathews, J. H. et al. (1984, supra)); Kimuro-Kuroda and Yasui, J Immunol (1988) 141:3603-3610).
Although the primary amino acid sequence of the flavivirus E glycoprotein is variable (45-80% identity), all have twelve conserved cysteine residues, forming six disulfide bridges, and hydrophilicity profiles are nearly superimposable, suggesting that they may all have similar secondary and tertiary structures. Based on the position of the 12 conserved cysteines (determined for West Nile virus, Nowak and Wengler, Virology (1987) 156:127-137), monoclonal antibody competitive binding studies, monoclonal antibody binding to purified proteolytic fragments, and analysis of neutralizing antibody escape mutants of Tick-Borne Encephalitis Virus, glycoprotein E was divided into three antigenic domains (A, B, and C) and two transmembrane segments at its carboxy-terminus. See, for example, Mandl, C. W. et al. J Virol (1989) 63:564-571. FIG. 1, reproduced from this article, shows the locations of these domains.
Domain A was defined as a denaturation sensitive, highly folded, and disulfide stabilized discontinuous domain composed of the amino acids from 50-125 and 200-250 containing five of the six disulfide bridges. Neutralization and hemagglutination inhibition epitopes are found within domain A, and, for dengue viruses, one of the two N-linked glycosylation sites. A conserved hydrophobic amino acid sequence within domain A has been postulated to provide fusogenic properties after low pH treatment. Amino acid sequences conserved among the flavivirus family are located within this region; thus, broadly flavivirus-cross-reactive epitopes lie within this domain.
Domain B was identified as a continuous domain composed of amino acids 301-395 (an approximate region between amino acids 300-400 for all flaviviruses). The domain can be isolated as a single immunoreactive proteolytic fragment. It has been postulated that this domain forms part of the receptor binding site (Heinz, F. X. et al. APMIS (1993) 101:735-745), and attenuating mutations have been mapped to sequences within domain B Heinz et al. (supra). A variety of neutralizing antibodies have been shown to specifically map to Domain B (Heinz et al. (1983, supra)); Trirawatanapong et al., 1992; Megret et al., 1992; Lin et al., 1994). The binding of these neutralizing monoclonal antibodies is dependent on formation of a disulfide bond, and in some cases also is sensitive to detergent denaturation. Species-specific monoclonal antibodies bind this domain.
Domain C represents a hypervariable loop between the two segments of Domain A. Its antigenicity is insensitive to denaturation or reducing agents, and contains one N-linked glycosylation site. Predominantly sub-type specific monoclonal antibodies react with this domain. No specific activity has been assigned to this domain.
Many strategies are currently under investigation to develop an effective and safe dengue vaccine; however, to date, no single strategy has proven completely satisfactory. Attempts to generate live attenuated dengue vaccine strains have not been entirely successful, although research into this area continues. In the absence of effective, live attenuated dengue vaccines, a significant effort has been invested in the development of recombinant, dengue subunit or viral-vectored vaccines.
Recombinant dengue proteins have been expressed in several systems to date (see Putnak, R. A. (1994) Modern Vaccinology, E. Kurstak ed., Plenum Medical, New York, pp. 231-252, for review). Most efforts using Escherichia coli have yielded poor immunogen unable to elicit neutralizing antibodies. This may reflect non-native conformation of dengue proteins expressed in the bacteria and the necessity to process the viral proteins through the secretion pathway in order to form the proper disulfide bonds, glycosylate the proteins, or both.
Several reports have described vaccinia-flavivirus recombinants expressing envelope protein as part of a polyprotein (e.g. C-preM-E-NS1; [Dengue] Zhao, B. G. et al. J Virol (1987) 61:4019-4022; Deubel, V. et al. J Gen Virol (1988) 69:1921-1929; Bray, M. et al. J Virol (1991) 63:2853-2856; [YF] Hahn, Y. S. et al. Arch Virol (1990) 115:251-265), as a single protein (e.g. 100% E; [Dengue] Bray, M. et al., J Virol (1989) 63:2853-2856), or as polypeptides (e.g. 79% E-RKG; Men, R. et al. J Virol (1991) 65:1400-1407). The most successful recombinant vaccinia viruses, those capable of inducing neutralizing antibodies and protecting mice from virus challenge, were the which were secreted E extracellularly or accumulated E on the cell surface.
Men, R. et al. (1991, supra) described the recombinant production of various C-terminal truncations of the DEN-4 envelope protein using a recombinant Vaccinia virus vector and infecting mammalian CV1 cells. The results showed that the recombinants that contain greater than 79% of the coding sequence produced an intracellular protein that could be immunoprecipitated with anti-dengue virus antibodies contained in hyperimmune mouse ascitic fluid (HMAF). Although there was a reduced level of detection for protein which contained 79% of envelope or less, this did not appear to result from reduced production of the protein. It was also found that only truncations which contained 79% of E or less were secreted efficiently; E polypeptides equal to or larger than 81% E were not secreted efficiently.
Men et al. (1991, supra) constructed additional C-terminal truncations between 79% E and 81% E to map the amino acids responsible for the difference in secretion and immunoreactivity with HMAF of these two truncated E polypeptides. The results demonstrated that 79% E containing the additional tripeptide sequence RKG was also secreted. Although both 59% E and 79% E-RKG were secreted, only 79% E-RKG was detected at the cells' surface. The recombinant Vaccinia viruses containing various truncations were also used to immunize mice. Mice immunized with recombinants expressing 79% E-RKG or larger portions of the envelope protein were protected. However, except for 59% E, mice immunized with 79% E or a smaller product were only partially protected. The 59% E elicited high protection rates (&gt;90%) comparable to 79% E-RKG and larger C-terminal truncated E polypeptides. Protection correlated with binding to HMAF.
Combinations of immunogenic structural and nonstructural JE virus, DEN-1, DEN-2, and DEN-4 proteins have been expressed by baculovirus recombinants (Matsuura, Y. et al. Virology (1989) 173:674-682; Putnak, R. A. et al. Am J Trop Med Hyg (1991) 45:159-167; Deubel, V. et al. Virology (1991) 180:442-447). Baculovirus-expressed dengue and JE E glycoprotein elicited neutralizing antibodies, protected mice from a lethal dengue virus challenge, or both. In spite of these successes, the expression levels reported in baculovirus are low and the recombinant protein is less immunogenic than the viral protein (Putnak, R. A. et al. Am J Trop Med Hyg (1991) supra).
Research with purified polypeptides released by proteolysis of flavivirus envelope proteins, with recombinant polypeptides, and with synthetic peptides has indicated where protective epitopes may map. The isolated 9000 dalton domain B trypsin fragment from TBE virus spontaneously refolds and is stabilized by disulfide bridges (Winkler, G. et al. J Gen Virol (1987) 68:2239-2244). This disulfide stabilized fragment elicited neutralizing antibodies in mice (Heinz, F. X. et al. Virology (1984) 130:485-501). In contrast, a 28,000 dalton trypsin fragment from WN virus containing domain B sequences was unable to spontaneously refold and did not elicit neutralizing antibodies (Wengler and Wengler, 1989).
A cyanogen bromide-cleaved 8 kD fragment (amino acids 375-456) overlapping domain B from JE envelope protein was found to induce neutralizing antibodies in mice (Srivastava, A. K. et al. Acta Virol (1990) 34:228-238). Immunization of mice with a larger polypeptide (JE E amino acid 319 to NS1 amino acid 65) spanning the 8 kD peptide expressed in Escherichia coli as a fusion to protein A elicited neutralizing antibodies and protected mice from lethal virus challenge (Srivastava, A. K. et al. Microbiol Immunol (1991) 35:863-870). This polypeptide begins between the two cysteines within domain B, and, therefore, cannot form the stabilizing disulfide bond that earlier reports suggest is necessary for presentation of protective epitopes.
Immunization of mice with synthetic peptides corresponding to amino acids within domain B, aa 356-376 from MVE (Roehrig, J. T. et al. Virology (1989) 171:49-60) or aa 352-368 from DEN-2 virus (Roehrig, J. T. et al. Virology (1990) 177:668-675), elicited low levels of neutralizing antibodies in mice, suggesting the presence in domain B of a weak linear neutralizing epitope (Roehrig, J. T. et al. 1989 and 1990, supra).
Mason, P. W. et al. J Gen Virol (1990) 71:2107-2114 identified two domains of the DEN-1 envelope protein: domain I which includes amino acids 76-93 of the E protein and domain II (equivalent to domain B) which includes amino acids 293-402. These domains were identified from deletion analysis using recombinant fusion proteins expressed in E. coli and reacted with antiviral monoclonal antibodies. Recombinant fusion proteins containing E. coli trpE sequences fused to the envelope protein (amino acids 1 to 412) elicited antibodies in mice which reacted with a portion of the protein containing domain II.
In addition, Mason, P. W. et al. (J Gen Virol (1989) 70:2037-2049) expressed a collection of E. coli trpE fusion proteins to segments of JE virus envelope protein spanning domain B. The trpE fusion proteins containing the smallest JE E fragments that retained immunoreactivity with a panel of neutralizing monoclonal antibodies included amino -acid residues from methionine 303 through tryptophan 396. However, animals immunized with immunoreactive trpE fusion polypeptides did not produce neutralizing antibodies nor were they protected from lethal challenge.
Trirawatanapong, T. et al. Gene (1992) 116:139-150. prepared several truncated forms of dengue 2 envelope proteins in E. coli for epitope mapping, and mapped monoclonal antibody 3H5 to its corresponding epitope. This was first localized between amino acids 255 and 422. Targeted gene deletions in the plasmid constructs encoding the truncated proteins permitted mapping of the binding site to the 12 amino acids between positions 386 and 397. The mapping was apparently confirmed by the ability of a synthetic peptide containing E protein amino acids 386-397 to bind 3H5 specifically.
Megret, F. et 41. Virology (1992) 187:480-491 prepared 16 overlapping fragments of DEN-2 envelope protein as trpE fusion products in E. coli for epitope mapping. The fusion proteins are produced intracellularly and obtained from the lysates. These products were used to map epitopes defined by a panel of 20 monoclonal antibodies. Six antigenic domains were described: non-neutralizing antibodies bound to peptides containing amino acids 22-58, amino acids 304-332, amino acids 60-97, and amino acids 298-397. Neutralizing antibodies bound to peptides containing amino acids 60-135, 60-205, and 298-397.
Significantly, Megret et al. (1992, supra) demonstrated that all MAbs (including 3H5), with two exceptions (below), that recognize "full-length" domain B (amino acids 298-397) are unable to recognize slightly shorter polypeptides. For example, in contrast to the findings of Trirawatanapong et al. Gene (1992, supra), MAb 3H5 was unable to bind to trpE fusion proteins containing DEN-2 E amino acids 304-397, 298-385, or 366-424. The two exceptional MAbs in the findings of Megret et al. are MAbs 5A2 and 9D12. The pattern of binding of MAb 5A2 suggests that it recognizes a linear epitope between amino acids 304 to 332, while MAb 9D12 binds to a polypeptide, amino acids 298-385, which is slightly shorter than the smallest polypeptide (amino acids 298-397) to which MAb 3H5 binds. These results indicate that both the disulfide bond in domain B and the domain B C-terminal amino acids are involved in forming the immunodominant domain B epitopes.
Although it appears established from the art that the B domain of the flavivirus envelope protein contains epitopes which bind neutralizing antibodies, problems have arisen with respect to producing recombinant polypeptides containing the B domain in a form which mimics the native protein and is thus capable of eliciting an immune response. The only recombinantly produced E polypeptides containing the B domain that elicited a protective immune response in mice were expressed from Vaccinia and baculovirus vectors. Generally, recombinantly produced proteins lack the appropriate 100-120 amino acid peptides, wherein the extensions do not interfere with the immunogenic effectiveness or secretion of the B domain. In one embodiment, such extensions are minimal--i.e., not more than six additional amino acids--at either the N-terminus or the C-terminus, or distributed between these termini; preferably no more than four total additional amino acids, and most preferably no more than two.
The form of domain B which spans positions of about 296-395 is referred to herein as "classical" domain B. When the B domain includes at least portions of the region extending to amino acid 413, the additional region may confer additional functions, e.g., enhancing immunogenicity by providing a helper T cell epitope. The form of domain B which includes positions about 296-413 is referred to herein as DomB+T. The domain B of the invention includes these two specific embodiments; "classical" domain B and DomB+T, as well as those forms which span positions approximately 296 to position between position 395 and 413.
Other portions of the E protein illustrated below are self-explanatory. 80% E is the N-terminal 80% of the protein from residue 1 to residue 395. 60% E represents the corresponding shorter sequence. These subunits are produced from vectors containing the DNA encoding the mature protein, or along with the prM fusion which results in secretion of the 80% or 60% per se.
For practical large-scale production of the subunits used as active ingredients in the vaccine of the invention, recombinant techniques provide the most practical approach. However, to be useful as active glycosylation, folding, and disulfide bond formation for producing a proper immune response.
It has now been found that the B domain of the envelope protein can be successfully secreted from yeast in a form which elicits the production of neutralizing antibodies. This permits, for the first time, the production of a successful recombinantly produced subunit dengue vaccine.